StemDeck, a free and open-source stem separation app, splits songs into up to six isolated tracks including vocals, drums, bass, guitar and piano, according to its GitHub repository. It is released under the Apache 2.0 license.
The app runs Demucs, Meta AI's neural network model for source separation, entirely on the user's own machine, so audio files never leave the device and no account is required, according to the repository. It accepts local files in formats including MP3, WAV, FLAC and M4A, as well as YouTube URLs.
Beyond separation, StemDeck includes a DAW-style waveform editor with mixer controls, VU meters, loop points, BPM detection, key analysis and loudness measurement, the repository states. It is built with a Python and FastAPI backend and ships as native desktop apps for macOS and Windows through Tauri, alongside a self-hostable web server version.
The project drew more than 100 points on Hacker News within a day of posting. Its local-only, no-account design sets it apart from subscription stem-separation services, giving musicians and producers a free alternative that keeps their audio off someone else's server.